Improved Forecasting Through the Design of Homogeneous Groups

The purpose of this paper is both to discuss the need to disaggregate economic data into meaningful groups in order to better understand and forecast the future course of economic phenomena, and to illustrate with a specific example that such disaggregation can lead to improved results. The reasons for placing observations into homogeneous groups has already been documented by the authors but will be reviewed briefly in the first section of this paper.' The next section will be concerned with the general procedure for grouping observations. The remainder of the paper will discuss in some detail the improvement in forecasting ability that comes from a specific application of grouping procedures to the problem of forecasting earnings per share for a large group of manufacturing concerns. Forecasts prepared on the basis of statistically grouped data will be compared with forecasts made on data grouped on traditional industrial criteria as well as with forecasts prepared by mechanical extrapolation techniques.